Nvidia's Fears Ease: A "Perfect Score" Earnings Report After the CEO Shakeup
After founder and CEO Jensen Huang stepped down amid a deepening controversy, Nvidia has temporarily eased the market's worries about the company's lack of a leader at the helm and about the sustainability of AI investment, once again delivering results that beat expectations on an already high base — and it raised its annual revenue guidance for the following fiscal year.
After the U.S. market close on Tuesday the 17th, Eastern time, Nvidia reported that for its fiscal 2027 third quarter ended October 25, 2026, revenue rose 103.1% year over year to $115.8 billion — not only the first time in the company's history that quarterly revenue has topped $100 billion, but also about 5.9% above analyst expectations and clearly above the company's own guidance midpoint of $108 billion.
Profitability was equally strong. Third-quarter non-GAAP adjusted earnings per share (EPS) rose 111% year over year to $2.68, 8.1% above market expectations; adjusted gross margin was 74.9%, slightly below the prior fiscal quarter's 75.0% but above the roughly 74% expected by the market.
The core data center business remained the growth engine. Third-quarter data center revenue kept its year-over-year doubling, reaching $107.6 billion, 5.6% above market expectations, and still accounts for roughly 90% of total company revenue. Compute and networking revenue was $106 billion, 7.9% above expectations; Graphics revenue was $9.8 billion, 10.9% above expectations.
Nvidia Chief Financial Officer (CFO) Colette Kress, who also serves as interim CEO, specifically cited key data center product Vera Rubin, saying it "got off to an explosive start, our product roadmap is on track, and our ambitions have never been greater."
More importantly, Nvidia's revenue guidance for the fourth quarter again came in stronger than market expectations: it expects fourth-quarter revenue with a midpoint of $130 billion, easily setting another record for a single quarter, 3.8% above the analyst consensus, higher than all buy-side expectations and in line with the highest sell-side estimate — the $130 billion from UBS. As with the third quarter, Nvidia noted that this quarter's guidance does not include data center compute revenue from China.
After the report was released, Nvidia shares, which had closed down more than 0.4% in Tuesday's regular session, moved higher in after-hours trading, with gains at one point reaching about 4%. On the earnings call, the board chairman made clear there was no intention to have Huang return to his post, and the stock gave back some of its gains. As Kress further raised the fiscal 2028 revenue growth target from more than 70% to about 75%, market sentiment improved again, and after-hours gains widened once more, rising more than 4%.
In the view of analysts, besides the upward revision to the fiscal 2028 revenue growth target, a number of other figures in this report also combined to ease market concerns. Nvidia's results for the past quarter and its guidance for the current quarter both grew more than expected, and although the gross margin guidance continues to decline sequentially, it is broadly in line with expectations and consistent with what Nvidia said on its second-quarter earnings call about gross margin having bottomed out for the time being. At the same time, Nvidia's statements about Vera Rubin shipments and its "fastest-ramping" product show that the new-generation platform has begun to enter the infrastructure buildout cycle of cloud service providers and AI cloud customers.
Overall, what Nvidia delivered this time is not just an earnings report with both revenue and profit above expectations, but also a strong bet on future AI infrastructure growth. Next, investors still need to watch whether Rubin mass production, gross margin recovery, customer payment capacity and long-term capital commitments can truly translate into sustained cash returns.
Revenue, EPS and gross margin all beat expectations; quarterly revenue tops $100 billion for the first time
Looking at the overall results, the core features of Nvidia's third quarter are: revenue above expectations, profit above expectations, gross margin above expectations, and guidance for the next fiscal quarter also above expectations.
Nvidia's third-quarter revenue was $115.8 billion, the first time a single fiscal quarter has topped $100 billion, up sharply from $96.2 billion in the second fiscal quarter, for sequential growth of about 20.3% and year-over-year growth of more than double. For comparison, Nvidia's second-quarter revenue grew 106% year over year, already a high growth rate for that stage; in the third quarter, even as the base continued to rise, the company still maintained year-over-year growth of more than 100%.
Earnings showed similar elasticity. On a non-GAAP basis, Nvidia's third-quarter net income was $64.87 billion, up 109% year over year and 20% sequentially; adjusted EPS was $2.68, up 111% year over year and 21% sequentially. On a GAAP basis, diluted EPS was $2.70, up 108% year over year.
Costs did not significantly erode profit. Third-quarter non-GAAP operating expenses were $8.84 billion, up 52% year over year, below revenue's 103% year-over-year growth; non-GAAP operating income reached $77.88 billion, up 116% year over year. That means Nvidia is still enjoying a clear scale effect driven by AI computing demand.
It is worth noting that starting from the first quarter of fiscal 2027, Nvidia's non-GAAP financial metrics no longer exclude stock-based compensation expense, and the company has restated historical non-GAAP data. As a result, this quarter's adjusted EPS and margins are now measured in a way that is closer to the true cost structure than before.
Data center revenue $5.7 billion above expectations, with both hyperscalers and AI cloud demand strong
The data center business remains the most critical part of the report. Third-quarter data center revenue was $107.6 billion, up 21% sequentially and 110% year over year, about $5.7 billion above market expectations.
In terms of composition, hyperscale cloud service providers are the main component: Nvidia's third-quarter Hyperscale revenue was $60.1 billion, while AI cloud, industrial and enterprise-related revenue was $47.5 billion. Together these two form the main source of data center revenue, and also show that AI infrastructure demand does not come only from a handful of internet giants, but extends to more scenarios such as AI cloud, industrial, enterprise and sovereign AI.
In terms of customer structure, hyperscalers remain the main force of demand, contributing $60.1 billion of revenue in the third quarter, up 71% year over year; AI cloud, industrial and enterprise customer-related business revenue reached $47.5 billion, up 195% year over year.
This set of figures shows that Nvidia's growth drivers are expanding further from traditional large cloud computing companies to AI-native companies, specialized AI cloud providers, and industrial and enterprise customers.
Recently, market concerns about the return on AI infrastructure investment have heated up: tech giants such as Microsoft, Google, Amazon and Meta continue to expand capital expenditure, while some emerging AI cloud providers face questions about financing capacity and cash flow. With data center revenue once again well above expectations, Nvidia shows, at least at the level of chip sales and revenue recognition, that demand for AI compute purchases has not yet clearly cooled.
It is worth noting that Nvidia's edge computing business also performed strongly, with third-quarter revenue of $8.2 billion, up 42% year over year and 14% sequentially, above the market expectation of $7.5 billion. Although far smaller than data centers, its growth reflects incremental contributions from on-device and edge applications such as AI PCs, local AI, robotics, autonomous driving and AI-RAN.
The company said this growth was driven by strong sales of its Blackwell workstation products as well as edge computing demand from Physical AI applications. At the same time, rising memory and system prices are still affecting consumer PC demand.
Looking at the overall business structure, AI demand is no longer confined to large data centers; it is also expanding into enterprise computing, edge devices, robotics and other physical AI scenarios.
Vera Rubin begins shipping, and Nvidia says it is the fastest product ramp in its history
Beyond the financial figures, the most important product signal from this report is that Vera Rubin NVL72 has begun shipping.
Nvidia says Vera Rubin NVL72 is the fastest-ramping product in the company's history, and related racks are already running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius, and are about to go live at SpaceXSI.
This directly addresses one of the market's biggest questions ahead of the report: whether the transition to the new generation of products after Blackwell is going smoothly, and whether Rubin can ramp supply as planned.
Nvidia says Vera Rubin NVL72 is the fastest-ramping product in its history. Currently, the related racks are already running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud and Nebius, and are planned to go live at SpaceXSI soon.
The company also announced an expansion of the Vera Rubin cloud ecosystem:
- Oracle Cloud launched Vera Rubin NVL72 systems and Vera CPU.
- Google Cloud launched A5X instances powered by Vera Rubin.
- Companies including Cisco, Google, Meta and Microsoft joined Vera Rubin infrastructure buildout.
- AI companies and cloud service providers such as CoreWeave, Meta, Nebius, Nscale and OpenAI are also building compute infrastructure based on NVIDIA Vera CPU.
Nvidia also announced it will expand key manufacturing capacity with supply chain partners including Amkor, IBIDEN, King Yuan Electronics, Samsung Electro-Mechanics, SPIL and Unimicron to meet future infrastructure demand.
This is particularly important for the market. Investors had been focused on whether the product transition after Blackwell would go smoothly, and whether Rubin could expand supply as planned. Now that Nvidia has confirmed the new-generation platform has begun shipping, at least on product progress some concerns have been eased.
In addition, Nvidia announced a rack-scale platform for the Vera CPU aimed at high-performance computing, as well as Rubin HPC GPUs designed for scientific computing, engineering simulation and data processing, seeking to extend the application scope of AI infrastructure further into scientific research and high-performance computing.
Fourth-quarter revenue guidance again above expectations, fiscal 2028 growth target raised to 75%
Compared with the above-expectations growth already achieved in the third quarter, Nvidia's outlook for future performance is even more closely watched.
The company expects fiscal 2027 fourth-quarter revenue of $130 billion, plus or minus 3%, implying a range of roughly $127.4 billion to $132.6 billion.
The guidance midpoint is nearly 4% above the roughly $125.25 billion analyst estimate compiled by Bloomberg, and also implies revenue growth of about 12% sequentially from the third quarter.
It is worth noting that the fourth quarter includes an extra week, so the sequential growth rate cannot be compared entirely on the basis of a normal quarter.
Another detail worth noting is that Nvidia explicitly said the fourth-quarter revenue outlook does not factor in any data center compute revenue from China. That means the company still expects to achieve further growth without relying on this portion of revenue.
More importantly, on the earnings call, Nvidia further raised its fiscal 2028 revenue growth expectation.
Interim CEO Kress said the company now expects revenue growth of about 75% year over year for fiscal 2028, which ends in January 2028, above its earlier target of more than 70%.
She attributed the upward revision to multiple factors, including a higher revenue base for this fiscal year than previously modeled, strong computing demand from hyperscalers, frontier AI labs, AI-native companies, ordinary enterprises and government customers, and additional support from supply chain partners.
This adjustment is particularly important. Previously, the market worried that Huang's sudden resignation might lead Nvidia to avoid long-term performance outlooks in this report, or even withdraw the growth target it had previously set. But the company not only maintained its original framework, it also raised expectations further.
According to Business Insider's coverage of the earnings call, Nvidia also expects frontier AI labs to contribute about 15% of its data center revenue.
That means frontier model developers such as OpenAI and Anthropic are becoming an increasingly important customer group for Nvidia's data center business.
However, the market still needs to watch whether such a high growth target can be consistently delivered, and whether the company can maintain profitability while rapidly expanding product supply.
Gross margin expected to fall to 72%, with memory costs and the new product ramp creating pressure
Despite the strong revenue guidance, Nvidia's outlook for fourth-quarter gross margin is relatively cautious.
The company expects fourth-quarter gross margin of 71% to 72% on both a GAAP and non-GAAP basis, clearly below the 74.8% and 74.9% of the third quarter.
That implies gross margin may decline by about 3 percentage points sequentially.
The gross margin pressure is mainly related to two factors: first, rising memory prices; second, the new-generation Rubin products are in the early ramp stage, with relatively high production and supply chain costs.
In addition, the company expects fourth-quarter GAAP operating expenses of about $11 billion and non-GAAP operating expenses of about $10.7 billion, also higher than some analysts had previously expected.
So Nvidia's current growth does not come without cost. As product iteration accelerates and R&D and infrastructure investment increase, the company needs to strike a balance between meeting customer demand and maintaining profit margins.
For investors, the next things to watch are not only whether revenue can continue to beat expectations, but also whether gross margin can gradually recover after Rubin ships at scale.
Supply commitments rise to $390 billion, plus $118 billion of guarantee exposure
There is another detail in this report worth a deeper look: Nvidia is deeply involved in AI infrastructure buildout through long-term procurement, cloud service agreements, data center leases, equity investments and financing guarantees.
According to the CFO commentary document, Nvidia's supply commitments as of the end of the third quarter increased to $390 billion from $279 billion in the prior fiscal quarter, mainly related to multi-year memory purchases.
In addition, the company's cloud service agreements and data center lease commitments totaled $102 billion, up from $74 billion in the previous quarter.
If equity investments, capital expenditure and other items are further included, Nvidia disclosed total future commitments of $518 billion.
Nvidia's future commitments as of October 25, 2026
Item Amount Supply and capacity commitments $390 billion Cloud service agreements $28 billion Data center leases not yet started $74 billion Equity investments $20 billion Capital expenditure $6 billion Total $518 billionThis set of figures shows that Nvidia is not only selling AI chips, but also locking in production capacity and business demand for years ahead through long-term supply contracts and infrastructure arrangements.
At the same time, the company disclosed third-party financing guarantee exposure of up to $118 billion.
The largest of these involves a data center campus being built by SB Energy in Ohio. Nvidia's related guarantee is for up to $105 billion, and the project will build about 4.25 gigawatts of data center capacity and provide infrastructure for OpenAI through a lease arrangement of up to 20 years.
However, these guarantees are not one-off payment obligations; they take effect in stages as the relevant buildings are completed and put into use. Nvidia said it has not made any payments under these guarantees to date.
The company also disclosed commitments related to emerging cloud service providers (Neoclouds) of $32 billion. Interim CEO Kress said Nvidia will continue to support this ecosystem, but in a more prudent and selective way.
She noted that some emerging cloud providers and frontier AI labs are growing faster than their balance sheets and long-term credit profiles can support. Nvidia will focus its support on projects with clear, sustainable demand.
This also reveals a change taking place in the AI supply chain: chipmakers are no longer just selling hardware to customers, but are gradually participating in the expansion of the entire AI compute ecosystem through supply commitments, infrastructure buildout and financing support.
But this model also means Nvidia has taken on more risk related to customer credit, long-term leases and capital allocation.
Receivables and inventory hit new highs, while capital returns continue to expand
Rapid growth has also brought the problem of increased working capital use.
As of the end of the third quarter, Nvidia's accounts receivable reached $81 billion, rising further from the prior quarter, with days sales outstanding (DSO) increasing from 60 days to 64 days.
The company explained this was related to the shipping pace after the Rubin product launch and to extended payment terms in multi-year agreements with some large customers.
In the same period, inventory reached $39.7 billion, up from $31.6 billion in the previous quarter, mainly to support the Rubin product ramp.
On cash flow, Nvidia's third-quarter operating cash flow reached $48.3 billion and free cash flow was about $43.8 billion, still maintaining strong cash generation.
The company also returned $28.1 billion to shareholders through share buybacks and cash dividends in the third quarter. As of the end of the quarter, remaining share buyback authorization was about $227.3 billion.
It is worth noting that the board's decision to increase share buyback authorization by $150 billion was actually approved in September 2026 and is not a buyback plan newly announced with this report.
In addition, Nvidia disclosed a $1.8 billion impairment on non-marketable equity investments in the third quarter. This also shows that as the scale of investment across the AI supply chain expands rapidly, valuation volatility and return risks on those investments are likewise worth watching.
Join the official Coincamps community:
Telegram: https://t.me/coin_camps
Recommended
Cardano Ecosystem Stablecoin DJED Announces Shutdown: After Nearly Four Years of Operation, Holders Must Complete Redemption Before New Year's Day
Nov 18, 14:45
From Crypto Venture Capital to Traditional Venture Capital: The Three Core Competencies That Venture Capital Actually Needs
Nov 18, 14:41
Nvidia Earnings Call: Board Rules Out Huang Return, Guides Q4 to $130B
Nov 18, 14:31
The AI bubble has already burst, yet retail investors are still frantically buying the dip: Hedge fund veteran warns Nvidia risks are far from cleared
Nov 18, 14:31
Research Report Analysis | Space Fund Size Grows to $18 Billion in One Year, but On-Orbit Direct Investment Accounts for Only 3%
Nov 18, 14:18
World Bank Partners with Citi and Circle: Global Aid Funds Expected to Be Settled On-Chain via USDC
Nov 18, 14:15